Research letter from the Evolving Flies of Switzerland
(September 29th 2008) New evolutionary insight: 'get smart and die young' or 'live long and stay dumb' It was once argued that just as the existence of a watch necessitates belief in an intelligent watchmaker, so the complexity of living organisms is evidence for the existence of a divine creator. In response, the evolutionary biologist Richard Dawkins described evolutionary processes as being analogous to a blind watchmaker. He used his computer program, The Blind Watchmaker, to illustrate the difference between potential complexity developing out of pure randomness and randomness combined with cumulative selection. By our corresponding author, Johannes Swatch
New evolutionary insight: 'get smart and die young' or 'live long and stay dumb' It was once argued that just as the existence of a watch necessitates belief in an intelligent watchmaker, so the complexity of living organisms is evidence for the existence of a divine creator. In response, the evolutionary biologist Richard Dawkins described evolutionary processes as being analogous to a blind watchmaker. He used his computer program, The Blind Watchmaker, to illustrate the difference between potential complexity developing out of pure randomness and randomness combined with cumulative selection.
In Switzerland, a country traditionally renowned for its intelligent watchmaking, evolutionary biologists are doing some blind watchmaking of their own to give evolution a helping hand. Heavily selecting for particular population traits in fruit flies is, they claim, providing experimental support demonstrating significant evolutionary associations between distinct phenotypes.
Dr Tadeusz Kawecki, from the University of Fribourg, has now reported a link between "Learning ability and longevity: a symmetrical evolutionary trade-off in Drosophila", in the journal, Evolution (2008; 62, 1294-1304). Here, he affirms that fruit flies that learn well die young and that flies that live longer aren't very smart.
Oddly, like Richard Dawkins, Kawecki's paper starts off with a machine analogy; in this case, of a robot, "RunBot is a biped robot that can adapt its walking pattern to terrain changes". Quite what this has to do with the study of flies is not immediately obvious, but it does give him the opportunity to explain that, "RunBot illustrates the adaptive power of learning". We are then assured that living organisms can also benefit from, "the positive effects of learning on fitness," but he then spends the rest of his article arguing that, on the whole, better learning really isn't very good for evolutionary fitness.
As Kawecki recently said in a New York Times article, "If it's so great to be smart, why have most animals remained dumb?"
His key argument is that, "learning ability has been substantially improved by artificial selection" of lab rats, honeybees and, thanks to his previous work, fruit flies. This suggests that, on an evolutionary scale, some species "have not fully realized the genetic potential for learning." Kawecki's hypothesis is that natural selection favours intermediate levels of learning ability because further improvements would be too costly.
Over a timescale of millions of years, many species could no doubt become better learners. But have Kawecki's selected fly populations really "evolved markedly improved learning ability" in the lab? And do they really drop dead when worn out by this extra cerebral strain?
High fliers place their eggs better
Kawecki's learning procedure, the oviposition learning test, only tests females. The flies are offered a choice of orange or pineapple jelly to eat and lay eggs on. During a 3-hour training period, one of the fruit jellies is laced with bitter-tasting quinine and the flies learn to avoid this unpleasant flavour. In the subsequent 3-hour testing period, the females again have the choice of orange or pineapple (this time without any quinine). But will they remember to avoid the fruit associated with quinine during their prior training period? The mothers that do remember lay their eggs on the correct fruit. To select for better learning as a population trait, Kawecki simply collects these eggs, then repeats the learning assay on the flies when they hatch. "Within 30 generations, the high-learning populations evolved substantially better performances", he says. But just how good are his smart flies? Even after 163 generations, the high-learners only score 0.33 on a learning index where 0 = no learning, and 1 = total learning. The control population scores just 0.2. However, the learning index is in fact a derivation. "No learning" in this assay should result in a 50:50 random distribution of eggs, so in fact only 10% of the control population and 16% of the selected "high-learners" appear correctly to have learned to associate bitterness with one of the fruit jellies. The large majority of the selected eggs are, apparently, still laid at random.
Clearly, even heavily selected fruit flies are not yet ready to inherit the earth. Especially when we learn that even this rather modest performance has such deleterious consequences for their health, "flies from high-learning populations lived significantly shorter than flies from control populations". The median life span in females was down 15% from 54 to 46 days and in males down 10%, from 48 to 44. However, although Kawecki says that there were no other differences in life history traits, his high-learners were "about 5% heavier in both sexes". But he dismisses this effect as "only marginally significant[1]". Nevertheless, his selected fly population is still bigger [2]- might this not also be due to their extra learning capacity?
An aged but dim population
Perhaps aware that his was not the strongest argument for a link between learning and longevity, Kawecki decided to do the test the other way round. He obtained two selected fly populations with an increased life span from Robert Arking in the US, 'La' and 'Lb', and put them through the oviposition learning test.
The populations were selected for late-life reproduction. Eggs were selected that had been laid by older mothers. Looking up Arking's work to find out why delayed female fecundity should result in longevity indicates that it is probably due to, "a delayed onset of senescence." In Kawecki's lab, these flies had a median life-span of 84 days compared to 66 days in the unselected, paired control population.
Well, Kawecki gave these long-lived ladies his oviposition learning test and says that they scored badly, displaying poor learning. But his data raises more questions. The flies were tested at the ages of 10, 20 and 30 days. At 10 days, the control flies had a learning index of 0.33 and the long-lived females just 0.21. But at 30 days both of them scored 0.2. Admittedly, these are entirely different populations. But at 10 days old the control population scores as well as the 163rd generation of his selected "high-learners", while the long-lived females have a score resembling that of the original control population. Overall, Drosophila don't seem to be very good learners.
However, Kawecki is convinced that, "fly populations selected for improved learning lived shorter [lives]" and, "fly populations selected for extended longevity had reduced learning ability early in life". Furthermore, he says, "these results are consistent with Williams's ninth prediction that 'successful selection for increased longevity should result in decreased vigour in youth'". So there!
"We don't know what the mechanism of this is," Kawecki told the New York Times, but "forming neuron connections may cause harmful side effects." He also suggested that it is worth investigating whether humans also pay hidden costs for extreme learning. "We could speculate that some diseases are a by-product of intelligence." Uh-oh!
It is interesting to compare Kawecki's confidence with the findings of Arking, who gave him the long-lived 'L' fly populations. Arking reported that his longer-lived flies display a significant resistance to oxidative stress due to enhanced expression of anti-oxidant genes and enzymes, but subsequently discovered that, although 'La' and 'Lb' are paired populations, selected in the 1980's under the same conditions from the same original wild population, they already display differences at the biochemical level, with markedly different increases in various anti-oxidant genes. To which Arking just shrugs, "the phenotypic equivalence observed at the organism level need not hold at the molecular genetic level". Fine, but just how confident can we be when looking at shifting traits in fly populations that may have complex genetic and environmental inputs? Perhaps it depends on how clearly the trait is defined and selected. For example, learning.
